XiaomiMiMo VL
MiMo VL 7B RL 2508
RL-2508 updates the earlier MiMo VL reinforcement-learning release. Xiaomi reports changes in image and video performance and adds an explicit way to turn reasoning off.
Check that the chosen mode is actually used
Practical guidance. A mode flag should be part of the evaluation record. Inspect both whether the model obeys it and whether the final answer remains correct. Internal preference results are useful context, but an application needs its own image and video examples.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | MiMo-VL-7B-RL-2508 |
| Licence | MIT.MiMo-VL-7B-RL-2508 official model card |
| Control | Thinking is the default; the documented no_think option requests a direct response.MiMo-VL-7B-RL-2508 official model card |
| Scope | The release compares the updated checkpoint with the earlier MiMo-VL-7B-RL, including an internal visual-model arena.MiMo-VL-7B-RL-2508 official model card |
Related models
- MiMo VL 7B RL
Xiaomi · MiMo VL
MiMo VL 7B RL continues the supervised vision-language checkpoint with mixed on-policy reinforcement learning.
Launch date not confirmedDownloadable - MiMo VL 7B SFT 2508
Xiaomi · MiMo VL
An updated MiMo VL 7B instruction model with a selectable reasoning mode.
Launch date not confirmedResearch releaseDownloadable - MiMo VL Miloco 7B
Xiaomi · MiMo VL
Miloco adapts MiMo VL 7B for home activities and hand gestures.
Launch date not confirmedRestricted accessDownloadable